Re: [PATCH net] net: let skb_orphan_partial wake-up waiters.
From: Eric Dumazet <edumazet@google.com>
Date: 2021-03-30 14:40:25
On Tue, Mar 30, 2021 at 4:25 PM Paolo Abeni [off-list ref] wrote:
quoted hunk ↗ jump to hunk
Currently the mentioned helper can end-up freeing the socket wmem without waking-up any processes waiting for more write memory. If the partially orphaned skb is attached to an UDP (or raw) socket, the lack of wake-up can hang the user-space. Address the issue invoking the write_space callback after releasing the memory, if the old skb destructor requires that. Fixes: f6ba8d33cfbb ("netem: fix skb_orphan_partial()") Signed-off-by: Paolo Abeni <pabeni@redhat.com> --- net/core/sock.c | 2 ++ 1 file changed, 2 insertions(+)diff --git a/net/core/sock.c b/net/core/sock.c index 0ed98f20448a2..7a38332d748e7 100644 --- a/net/core/sock.c +++ b/net/core/sock.c@@ -2137,6 +2137,8 @@ void skb_orphan_partial(struct sk_buff *skb) if (refcount_inc_not_zero(&sk->sk_refcnt)) { WARN_ON(refcount_sub_and_test(skb->truesize, &sk->sk_wmem_alloc)); + if (skb->destructor == sock_wfree) + sk->sk_write_space(sk);
Interesting.
Why TCP is not a problem here ?
I would rather replace WARN_ON(refcount_sub_and_test(skb->truesize,
&sk->sk_wmem_alloc)) by :
skb_orphan(skb);
This will get rid of this suspect WARN_ON() at the same time ?
skb->destructor = sock_efree;
}
} else {
--
2.26.2